What happened
GitHub reported elevated errors and latency across Issues, Pull Requests, APIs, Actions and Copilot on August 17, 2026. Its official incident report records an incident window of 13:28–21:15 UTC: 7 hours, 47 minutes. Individual services recovered at different times.
Recovery timeline (UTC)
| Time on August 17 | Provider-reported milestone |
|---|---|
| 13:28 | Incident began |
| 16:36 | Most services recovered |
| Approximately 18:03 | Actions degradation ended |
| 21:02 | Copilot Token Service fully recovered |
| 21:15 | Overall incident resolved |
GitHub attributed the disruption to saturated load balancers in its Central US datacenter, involving a sidecar concurrency limit and an autoscaling configuration problem. Retries prolonged recovery, particularly for Copilot authentication.
What this means
The incident window does not mean every customer or feature was unavailable throughout. This is a summary of GitHub's account, not an independent measurement of customer downtime.
Correction and citation
Corrected September 12, 2026: the previous version's 6-hour, 45-minute duration and generic recovery timeline were unsupported. This revision uses GitHub's postmortem. When citing this brief, link here and retain the UTC times and provider attribution.
For current conditions, check GitHub status or GitHub Actions status.